Transcript Document 7628005

Taking the guesswork out of
author searching
- the why, the how & the what -
Niels Weertman
Head of Scopus Product Management
Annual CrossRef meeting
Cambridge – November 1st, 2006
Why?
 Finding author-related information is one of the most
common search patterns.
 Author searching in databases was hampered by two
serious problems:
 How to distinguish between an author’s articles and those of
another authors sharing the same name?
 How to group an author’s articles together when his or her
name has been recorded in different ways?
2
Some examples
 A Nobel laureate







Theodor Haensch
T. Haensch
Theodor W. Haensch
Theoder Hänsch
T. Hänsch
Theodor W. Hänsch
…
 Two different authors:
 J.R. Weertman and J.R
Weertman
 Material Science
 Northwestern University
 Inaccurate and incomplete results
 Time-consuming
3
How?
 Three possible approaches:
 User-created
 Authority file
 Algorithm
 How did we approach this?
 Top down approach: algorithm using array of data in records.
 Incorporating a ‘bottom up’ aspect by including author
feedback on where our matching or data needs improvement.
4
What information do we use?
Source title
Cited by
Author & co-author
5
Affiliation
What information do we use?
references
6
Dedicated author search page
7
One click and select your author
8
Author feedback
9
Only part of the whole story …
 Search is only a means to achieve something
else:
 Find new articles, citations, etc.
 Evaluate the author’s work
 Identify co-authors
 We also offer a dedicated author details page
where users can “jump to” specific features for
each author
10
Author “jump to” page
11
Reaction on author disambiguation
1,500+ e-mails
 In general authors are extremely positive and
understanding of errors
1,300+ corrections
 Each one has been reviewed for accuracy and
legitimacy, resulting in splits and merges
 The remainder under investigation our pending
outstanding action
12
Where to from here?
 Continuously improve precision and recall
 Improve data and precision and recall by adding
via author feedback and new or corrected data
 What users said they want …
 Even easier searching
 More options once the author is found
13
Any Questions?
[email protected]
14